Published on 06 January 2014. Downloaded by UNIVERSITY OF ALABAMA AT BIRMINGHAM on 30/10/2014 03:57:20.
PCCP View Article Online
PAPER
Cite this: Phys. Chem. Chem. Phys., 2014, 16, 4268
View Journal | View Issue
The role of charge transfer in the energy level alignment at the pentacene/C60 interface J. Beltra´n,* F. Flores and J. Ortega Understanding the mechanism of energy level alignment at organic–organic interfaces is a crucial line of research to optimize applications in organic electronics. We address this problem for the C60–pentacene interface by performing local-orbital Density Functional Theory (DFT) calculations, including the effect of the charging energies on the energy gap of both organic materials. The results are analyzed within the
Received 27th November 2013, Accepted 25th December 2013
induced density of interface states (IDIS) model. We find that the induced interface potential is in the
DOI: 10.1039/c3cp55004d
mainly induced by the small, but non-negligible, charge transfer between the two compounds and the
range of 0.06–0.10 eV, in good agreement with the experimental evidence, and that such potential is multipolar contribution associated with pentacene. We also suggest that an appropriate external
www.rsc.org/pccp
intercompound potential could create an insulator–metal transition at the interface.
I. Introduction Organic electronics has emerged as a promising field of research devoted to improving the performance of devices such as light emitting diodes, photovoltaic cells or field effect transistors. In these multilayered organic-based devices a suitable tailoring of properties like high photocurrent quantum efficiencies can be achieved using different combinations of electron-donor and electron-acceptor layers.1–5 For such donor–acceptor heterojunctions some of the considered constituting materials are C60, whose mechanical, electrical and chemical properties are favorable for the fabrication of reliable devices,6 and pentacene (P5) which shows a high mobility of holes7 and distinct polymorph phases.8 It is also well-known that in organic electronics the role of the interface is crucial since key processes such as charge transfer or charge recombination occur in its vicinity. Consequently, a great effort has been made to understand the electronic properties of donor–acceptor interfaces and their related phenomena such as band bending and interface energy level alignment. The interface energy level alignment at organic/organic interfaces depends on the interface structure which is partially controlled by the growth process. Using appropriate methods like Molecular Beam Epitaxy, abrupt interfaces have been prepared and, in those cases, it has been found that the band alignment of the organic materials forming the junction is controlled by an interface induced potential,9,10 at variance with the Schottky–Mott limit in which the vacuum levels of both compounds are assumed to be aligned. Several models, like the Integer Charge Transfer11 and the Induced Density of
Interface States (IDIS),10 have been proposed to explain the formation of that potential. In particular, the IDIS-model relies on parameters like the charge neutrality level (CNL), which is an electronegativity marker for organic materials that predicts the amplitude and direction of charge transfer at the interface, and the interface screening parameter (S) which dictates how efficiently a shift in the CNL affects the induced potential.10 On the other hand, it has also been proposed recently by Linares et al.12 that, for a C60/P5 cluster, the level alignment shift for those materials is mainly provided by the multipole electrostatic potential created by the organic molecules. In this paper, we analyze the electronic properties and interface energy level alignment for the P5/C60 interface including in our DFT calculations the effects associated with the molecule charging energy that will allow us to correct appropriately the transport energy gap of both organic materials.13–17 In our calculations, we go beyond cluster models, as used in other studies,12,18 and consider periodic geometries that yield a better description of the long range potential created on the organic molecules by the charge rearrangement induced at the interface.13 This will allow us to make a detailed comparison between the effects on the interface energy level alignment due to the charge transfer or to the molecule multipole potentials. The P5/C60 interface geometry is analyzed in Section II. Then, in Section III we discuss our approach to analyze the energy level alignment at the interface between two organic materials using DFT calculations. Finally, in Section IV we present our results and conclusions.
II. Interface geometry ´rica de la Materia Condensada and Condensed Matter Physics Depto. de Fı´sica Teo Center (IFIMAC), Universidad Auto´noma de Madrid, 28049-Madrid, Spain. E-mail: [email protected]
4268 | Phys. Chem. Chem. Phys., 2014, 16, 4268--4274
Regarding the interface structure, predictions continue to be a challenging task12 with several studies19,20 suggesting a
This journal is © the Owner Societies 2014
View Article Online
Published on 06 January 2014. Downloaded by UNIVERSITY OF ALABAMA AT BIRMINGHAM on 30/10/2014 03:57:20.
Paper
disordered interface. In view of this lack of reliable information we have proceeded to calculate two different interface geometries, P5(001)/C60 and P5(011% )/C60, see Fig. 1 and 2. In the P5(001)/C60 interface, P5 is mainly oriented perpendicular to the surface, while in the P5(011% )/C60 interface P5 is almost parallel. These two interfaces represent two limiting cases for the P5 orientation, a fact that will allow us to suggest general conclusions even for disordered interfaces. Thus, we have calculated for each interface the most plausible geometry using DFT calculations that incorporate van der Waals interactions in a semi-empirical fashion. In these calculations we have used the local-orbital DFT code FIREBALL.21–24 FIREBALL uses real-space techniques and a basis set of numerical atomic-like orbitals (NAOs).25,27 This technique is based on a local-orbital formulation of DFT in which self-consistency is implemented on the orbital occupation numbers (see ref. 21 for details); in particular, in the present calculations the orbital occupation ¨wdin numbers have been obtained using the orthonormal Lo orbitals.24 We have used an optimized NAO basis set of s-(H) and sp3-(C) orbitals, with the following cut-off radii (in a.u.): s = 4.1 for H, and s = 4.5, p = 4.5 for C,26 and a k-space grid of 32 points homogeneously distributed in the 2-dimensional Brillouin zone. In these calculations we have used the Local Density Approximation (LDA)22 and the ion–electron interaction is modeled by means of norm-conserving pseudopotentials.27 In order to provide a more reliable geometry for the P5/C60 interface, a van der Waals interaction energy is included as an atom–atom attractive potential, fD(R)C6/R6, where R is the distance between each pair of atoms at the interface and fD(R) is the Grimme damping expression;14,28 C6 is taken as 14.5 eV A6 for the C–C interaction,29 which provides the relevant contribution to this energy.
Fig. 1 Adsorption geometry for the P5(011% )/C60 interface: top view and side view. The in-plane lattice vectors and the interface distance d are also shown. The two different C60-molecules are indicated (C60-1 and C60-2); and only the first P5 layer is plotted for simplicity.
This journal is © the Owner Societies 2014
PCCP
Fig. 2 Adsorption geometry for the P5(001)/C60 interface: top view and side view. The in-plane lattice vectors and the interface distance d are also shown.
It should be emphasized that, in order to avoid the double counting that might appear by combining simultaneously the LDA exchange–correlation energy and the van der Waals forces, we have introduced a kind of ‘‘corrected’’ LDA-approach13–15 whereby we eliminate the exchange–correlation contribution associated with the overlapping densities of both organic materials (see ref. 14 for more details). Then, the total interaction energy is calculated as the sum of the energies obtained using this ‘‘corrected’’ LDA-approach and the van der Waals forces as described above. P5 is known to have different polymorphs30 so its structure can vary depending on factors such as temperature or other growing conditions. We have considered the P5(001) and (011% ) surfaces at room temperature as given in references,30,31 and have combined each surface with a C60(001) structure with a distance between C60 neighboring centers approximately equal to 10 Å.32 Then, in order to obtain a plausible structure for the P5(011% )/C60 and the P5(001)/C60 interfaces, we have proceeded in the following way: (a) on the P5(011% ) surface structure, we have initially considered one C60 molecule in the unit cell (lattice vectors: a = 12.4 Å; b = 14.7 Å, see Fig. 1) letting the system relax until the atomic forces are smaller than 0.05 eV Å1. The obtained position, denoted C60-1 in Fig. 1, agrees well with one of the most stable configurations of a recent study.33 We then included a second C60 molecule in the unit cell (C60-2 in Fig. 1), relaxing all atomic positions again. We should stress that, with
Phys. Chem. Chem. Phys., 2014, 16, 4268--4274 | 4269
View Article Online
Published on 06 January 2014. Downloaded by UNIVERSITY OF ALABAMA AT BIRMINGHAM on 30/10/2014 03:57:20.
PCCP
Paper
the chosen P5 polymorph (see Fig. 1), the final C60–C60 distance at the interface is 9.6 Å, to be compared with 10 Å for the C60 bulk phase.32 Notice the different vertical distances of the two C60 positions on the top P5 surface: for the C60-1 molecule (blue circle in Fig. 1) the distance d from the layer of highest H atoms on P5 to the closest C atom in C60 is 1.7 Å, while for the C60-2 molecule that distance is 2.2 Å. (b) For the P5(001)–C60 interface, we only have one C60 molecule per unit cell; then, we again relax the C60 position until forces are smaller than 0.05 eV Å1, see Fig. 2. In this case the distance d is 2.0 Å, which is in-between the two values obtained for the P5(011% )/C60 interface. The in-plane lattice vectors in this interface form an angle of 102.31 and are 9.75 Å and 10.38 Å long while, as mentioned above, the C60–C60 distance in bulk C60 is 10.32
III. Calculation of energy levels at the organic/organic interface The theoretical analysis of the energy level alignment in organic/organic interfaces is a challenging task due the large size of these systems as well as the deficiencies of standard DFT techniques to properly describe the electronic structure of these organic interfaces.10 Due to the complexity of this analysis we use a practical and simplified approach13–15 based on local-orbital DFT (see Section II) in which appropriate corrections are introduced to take care of the main inaccuracies of the DFT calculation, see below. It is well known that in Kohn–Sham DFT calculations for organic semiconductors, local or semi-local exchange–correlation functionals give rise to small band gap values.10,17,34 In this respect, the critical role that different hybrid exchange–correlation functionals might play in the electronic properties of those materials at organic heterojunctions has been recently studied.35 In order to discuss how we have proceeded to tackle this problem at the P5/ C60 interface it is convenient to start presenting how accurate ionization and the affinity levels, eI and eA, for the isolated molecules can be determined from total energy DFT calculations for the neutral, anionic and cationic charged molecules respectively with N, N + 1 and N 1 electrons (D-SCF approach): eI = E [N] E [N 1] eA = E [N + 1] E [N],
(1)
where E [Ni] is the total energy for the ground state of the molecule with Ni electrons. The results of D-SCF calculations are usually in good agreement with the experimental evidence for the ionization and the affinity levels, eI and eA, for the isolated molecules. In contrast, the HOMO and LUMO Kohn– Sham levels obtained using local or semi-local exchange–correlation DFT functionals do not properly represent the transport gap for organic molecules. This is illustrated in Fig. 3. The experimental transport gap (ETg)0 = (eA eI), for C60 (P5) is 4.9 eV (5.2 eV) and the mid-gap is located 5.4 eV (4.0 eV) from the vacuum level. For comparison, the energy gaps obtained from converged36,37 DFT-LDA calculations, ELDA , for the neutral g molecules are: 1.6 eV (C60) and 1.2 eV (P5), very far from the
4270 | Phys. Chem. Chem. Phys., 2014, 16, 4268--4274
Fig. 3 (left) Energy level diagram showing the vacuum, highest occupied and lowest unoccupied levels as obtained from DFT Kohn–Sham LDA calculations for the isolated C60 and P5 molecules. (right) Experimental values of the affinity (eA) and ionization levels (eI) for the C60 and P5 molecules.
experimental values. This problem is related to the molecule self-interaction energy as described by the molecule charging energy U,10,17 i.e. (ETg)0 = ELDA + U. In Fig. 3 (left) we show the g LUMO and HOMO levels of the converged DFT-LDA calculations; notice also that the relative alignment of the DFT-LDA energy mid-gaps also differs from the ones obtained from experiment (or from a D-SCF calculation): while this difference is 0.8 eV in Fig. 3 (left), it amounts to 1.4 eV in Fig. 3 (right). The energy levels of the isolated molecules are shifted at the interface due to the interaction with the other molecules. The effect of this environment is threefold:17 (A) broadening of the energy levels into bands; (B) relative shift of the C60 and P5 molecular levels due to electrostatic and Pauli exclusion effects; (C) ‘‘dynamical’’ polarization or screening effects, shifting in different directions of occupied and empty states. Points (A) and (B) are obtained from our DFT calculations; in particular point (B) gives rise to an interface potential between both crystals. The effect of point (C) is a reduction of the transport gaps at the interface (or in the crystals) as compared with the values for isolated C60 or P5 molecules: ETg = (ETg)0 dU. Moreover, this gap reduction is also reflected in a shift of the ionization and affinity levels given by dU/2 and dU/2, respectively.10,17 The effects of the self-interaction energy and dynamical polarization response on the interface electronic structure are included in our calculations in a practical and simplified way introducing for each molecule a scissor operator:10 Ua X Oscissor ¼ (2) fjmi ihmi j jn i ihn i jg; a 2 ðmnÞ |mii and |nii being the empty (occupied) orbitals of the isolated molecule; Ua is the charging energy of the a-molecule. Notice that in eqn (2): U = (ETg)0 ELDA dU for each molecule a. We g also introduce for each molecule a rigid shift of the molecular levels by means of a shift operator: X Oshift ¼ ea (3) jbihbj; a ðbÞ
This journal is © the Owner Societies 2014
View Article Online
Published on 06 January 2014. Downloaded by UNIVERSITY OF ALABAMA AT BIRMINGHAM on 30/10/2014 03:57:20.
Paper
|bi being the orbital for each isolated molecule. Using these operators, we can fix in our local-orbital DFT calculations the initial value of the HOMO–LUMO gap and relative mid-gap position for the P5 and C60 molecules. Still, we have to determine the appropriate values of Ua and ea for each molecule at the P5/C60 interface. For this, instead of trying an ab initio approach we have resorted to using the experimental evidence.38–40 Several experimental studies38–40 have reported energy gaps for C60 and P5, respectively of 2.6 and 2.2 eV, at the P5/C60 interface. It is worth mentioning that theoretical calculations for the C60/Au(111) and P5/Au(111) interfaces14,16 have yielded energy gaps of 3.1 eV (C60) and 2.7 eV (P5). Notice that in both cases the energy gap for C60 is 0.4 eV larger than that for P5, the difference being that the experimental gaps for the P5/C60 interface are 0.5 eV smaller than the theoretical ones for the Au(111)/organic contact. Thus, the values ETg(C60) = 2.6 eV and ETg(P5) = 2.2 eV seem very reasonable. This information is completed with the values of the mid-gap positions for the P5 and C60 molecules, which are obtained from Fig. 3 (right). Thus, we use the scissor and shift operators, eqn (2) and (3), to adjust the initial values of the energy gaps and mid-gap positions for each molecule; then, our DFT calculation is performed after the operators, Oshift and a Oscissor , are introduced as a kind of pseudopotential. a
IV. Results and discussion Fig. 4 shows the energy level schemes for both the P5(011% )/C60 and P5(001)/C60 interfaces after electronic self-consistency. As shown in this figure, the interface induced potential at the P5(011% )/C60 interface is 0.14 eV, while for the P5(001)/C60 interface is 0.04 eV. The average of these two values, 0.09 eV, can be taken as a good representation of the interface potential for a disordered interface and agrees well with the experimental interface potential ranging between 0.07 eV and 0.15 eV.11,38,39
Fig. 4 Energy level diagrams of the vacuum, highest occupied and lowest unoccupied levels for the P5(011% )/C60 (left) and P5(001)/C60 (right) interfaces after electronic selfconsistency. These diagrams correspond to the D = 0 case (see text).
This journal is © the Owner Societies 2014
PCCP
For the (LUMOC60 – HOMOP5) difference we find 1.1 eV at the P5(011% )/C60 interface and 0.9 eV at the P5(001)/C60 interface, which should be compared with reported experimental values between 1.6 and 1.1 eV.11,38,39 A good understanding of the mechanism behind the interface induced potential can be reached by analyzing how the interface properties depend on the initial relative alignment between the mid-gaps of both materials; this would correspond to applying an external potential (D), or a rigid shift to, say, the P5 energy levels with respect to the C60 levels. This shift is introduced in our calculations using the Oshift operator mentioned a above, see eqn (2). In this way we can analyze the induced potential, the charge transfer and the (LUMOC60 – HOMOP5) evolution as a function of D. Notice that the cases shown in Fig. 4 correspond to D = 0 eV. Fig. 5 shows, for the P5(011% )/C60 and P5(001)/C60 interfaces, the (LUMOC60 – HOMOP5) and the charge transfer per molecule as a function of D (positive values of D means that the P5 levels are displaced upwards). For negative values of D, (LUMOC60 – HOMOP5) increases and the charge transfer is small. However, for positive values of D the difference (LUMOC60 – HOMOP5) decreases and the charge transfer from P5 to C60 increases, indicating that this charge transfer is associated with the interaction between the LUMOC60 and HOMOP5 levels. Interestingly, for large values of D (D > 0) the difference between the LUMOC60 and the HOMOP5 level for the P5(011% )/C60 and P5(001)/C60 interfaces converges to 0.50 and 0.25 eV respectively. This is related to the onset of a strong overlap between the LUMOC60 and HOMOP5 states, which is also reflected in the
Fig. 5 (LUMOC60 – HOMOP5) difference and charge transfer per molecule at the P5(011% )/C60 and P5(001)/C60 interfaces as a function of D (see text). The estimated charge error is around 0.005e.
Phys. Chem. Chem. Phys., 2014, 16, 4268--4274 | 4271
View Article Online
Published on 06 January 2014. Downloaded by UNIVERSITY OF ALABAMA AT BIRMINGHAM on 30/10/2014 03:57:20.
PCCP
Paper
Fig. 6 Density of states projected on the P5 or C60 organic materials for the P5(011% )/C60 interface. We have used a Lorentzian broadening of 0.05 eV for the sharp states. Left and right plots correspond to D = 0 and D = 1.5 eV respectively; notice the overlap between the HOMOP5 and the LUMOC60 levels in the latter.
significant increase of charge transfer as D increases further. This strong overlap gives rise to a metallic interface. Fig. 6 (left) shows the density of states for the P5(011% )/C60 interface projected onto either the P5 or C60 molecules, for D = 0. In this case the HOMOP5 level is around 1 eV below the LUMOC60 level and the charge transfer is small (see Fig. 5). Hybridization features, such as interfacial states corresponding to the interaction between P5 and C60, can be seen at the middle of the P5 band gap. For comparison, Fig. 6 (right) shows the density of states for D = +1.5 eV, a case presenting a metallic interface due to the strong overlap between the HOMOP5 and the LUMOC60 levels. These results show that the charge transfer plays an important role in the level alignment between the C60 and P5 organic layers. In particular, the potential induced between the organic compounds, Vinduced, follows the trend of the charge transfer. Fig. 7 shows this induced interface potential, Vinduced, as a function of D; notice how Vinduced increases for positive values of D as corresponds to a larger charge transfer between both materials, showing also a change of trend to a metallic interface for values of D around 0.5 eV. It is interesting to point out that for D > 1 eV the initial position of the P5 HOMO level is above the C60 LUMO level, suggesting a transfer of charge of 1 electron from P5 to C60. The transfer of charge between the P5 and C60 layers creates, however, a large long-range induced dipole potential between both layers (Fig. 7), shifting downwards the P5 levels w.r.t. the C60 levels. The final values for the P5 and C60 levels and charge transfer are obtained from the selfconsistent solution of this problem (Fig. 5). For example, for D = 2 eV this yields a charge transfer of 0.15–0.35 electrons per molecule and a final position for the P5 HOMO level 0.25–0.5 eV below the C60 LUMO level (Fig. 5). On the other hand, for negative values of D, Vinduced decreases quite slowly, as corresponds to an insulating interface. This situation can be described by means of the screening parameter S35 that can dVinduced (for a full insulating dD interface S = 1, i.e. no screening of the external potential D, while S = 0 for a completely metallic case). For the insulating be obtained from Fig. 7 as S ¼ 1
4272 | Phys. Chem. Chem. Phys., 2014, 16, 4268--4274
Fig. 7 Induced interface potential, Vinduced, as a function of D for the P5(011% )/C60 and P5(001)/C60 interfaces.
regime we find S = 0.96 0.02 while for the metallic one S = 0.03 0.01, this last small value reflecting the high screening properties associated with the metallic interface. We also find that the charge transfer vanishes (see Fig. 5) for D C 1.6 or +0.4 eV for the P5(011% )/C60 and P5(001)/C60 interfaces, respectively. For these values of D we find that Vinduced = 0.025 or 0.065 eV for the P5(011% )/ C60 and P5(001)/C60 interfaces, respectively, but not zero, indicating that another mechanism different from the charge transfer is operating here: as shown by Linares et al.12 this is associated with the potential created by the multipoles of the neutral P5 molecules, a mechanism responsible for the interface potential appearing at zero charge transfer. Therefore, in our calculations we find that for the P5(011% )/C60 interface, in the case D = 0, Vinduced = 0.14 eV while the multipolar contribution is 0.025 eV. However, for the P5(001)/ C60 interface (D = 0), Vinduced = 0.04 eV, the multipolar contribution being as large as 0.065 eV. Taking the average of these two interfaces as a reasonable representation of a disordered interface, we find that Vinduced C 0.09 eV, while the multipolar potential is B0.045 eV. We conclude that, in addition to the charge transfer mechanism, the multipolar term yields an important contribution to the total induced interface potential for D = 0, in agreement with ref. 12. Finally, we should stress that, as commented above, for large positive values of D we find a strong overlap between the
This journal is © the Owner Societies 2014
View Article Online
Published on 06 January 2014. Downloaded by UNIVERSITY OF ALABAMA AT BIRMINGHAM on 30/10/2014 03:57:20.
Paper
acceptor’s LUMOC60 and donor’s HOMOP5 levels, making the interface to have a metallic behavior and to show a pinning of the (LUMOC60 – HOMOP5) value. This indicates that a metal–insulator transition appears at this interface for D B 0.5 eV, associated with the overlapping of the LUMOC60 and the HOMOP5 levels (see Fig. 6 (right)). We also mention that a similar transition has been recently reported for the TCNQ/TTF interface;14,41 in that case, however, the TCNQ/TTF interface is already metallic for D = 0. For the P5–C60 case, we suggest that an appropriate external potential between the two organic materials could create that kind of metal–insulator phase transition. In conclusion, we have analyzed the energy level alignment at the interface between the C60 and pentacene organic semiconductors using a DFT approach with corrections to properly describe the energy gaps of the organic materials. We have studied two different interface geometries, corresponding to the (001) and (011% ) surfaces of pentacene. We find that the interface potential induced at the P5/C60 interface is mainly due to two effects: the charge transfer between the two materials and the multipolar contribution associated with pentacene. For a disordered P5/C60 interface we find, after averaging over the values calculated for the P5(011% )/C60 and P5(001)/C60 interfaces, that the total induced interface potential is B0.09 eV, while the multipolar contribution to that induced interface potential is found to be B0.045 eV. We have also shown that, depending on an effective external potential acting between the C60 and P5 organic materials, the interface can show either an insulating or a metallic character.
Acknowledgements This work is supported by the Spanish MICIIN under contract FIS2010-16046, the CAM under contract S2009/MAT-1467, the European Project MINOTOR (Grant FP7-NMP-228424) and COST-CMTS Action CM1002 (CODECS).
References 1 R. Cantrell and P. Clancy, Surf. Sci., 2008, 602, 3499–3505. 2 S. Duhm, G. Heimel, I. Salzmann, H. Glowatzki, R. L. Johnson, A. Vollmer, J. P. Rabe and N. Koch, Nat. Mater., 2008, 7, 326. 3 T. Hosokai, N. Mitsuo, S. Noro, T. Nakamura, S. Kera, K. Sakamoto and N. Ueno, Chem. Phys. Lett., 2010, 487, 67. 4 K. Akaike, K. Kanai, Y. Ouchi and K. Seki, Adv. Funct. Mater., 2010, 20, 715. 5 N. Hiroshiba, R. Hayakawa, T. Chikyow, Y. Yamashita, H. Yoshikawa, K. Kobayashi, K. Morimoto, K. Matsuishi and Y. Wakayama, Phys. Chem. Chem. Phys., 2011, 13, 6280. 6 D. Dougherty, W. Jin, W. Cullen, G. Dutton, J. Reutt-Robey and S. Robey, Phys. Rev. B: Condens. Matter Mater. Phys., 2008, 77, 073414. 7 K. Kim, Y. K. Yoon, M. Mun, S. P. Park, S. S. Kim, S. Im and J. H. Kim, J. Supercond., 2002, 15, 595–598. 8 C. C. Mattheus, G. a. de Wijs, R. a. de Groot and T. T. M. Palstra, J. Am. Chem. Soc., 2003, 125, 6323–6330.
This journal is © the Owner Societies 2014
PCCP
9 H. Ishii, K. Sugiyama, E. Ito and K. Seki, Adv. Mater., 1999, 11, 605; I. G. Hill, et al., Appl. Surf. Sci., 2000, 166, 354. ´zquez, Phys. Chem. Chem. Phys., 10 F. Flores, J. Ortega and H. Va 2009, 11, 8658–8675. 11 X. Liu, Y. Zhan, S. Braun, F. Li and M. Fahlman, Phys. Rev. B: Condens. Matter Mater. Phys., 2009, 80, 115401; S. Braun, W. R. Salaneck and M. Fahlman, Adv. Mater., 2009, 21, 1450. 12 M. Linares, D. Beljonne, J. Cornil, K. Lancaster, J. L. Bredas, S. Verlaak, A. Mityashin, P. Heremans, A. Fuchs, C. Lennartz, P. Aurel and L. Ducasse, J. Phys. Chem. C, 2010, 114, 3215–3224. ´n, F. Flores, J. I. Martı´nez and J. Ortega, J. Phys. 13 J. I. Beltra Chem. C, 2013, 117(8), 3888. 14 E. Abad, J. Ortega and F. Flores, Phys. Status Solidi A, 2012, 209, 636; E. Abad, J. Ortega, Y. J. Dappe and F. Flores, Appl. Phys. A: Mater. Sci. Process., 2009, 95, 119–124. 15 E. Abad, Y. J. Dappe, J. I. Martı´nez, F. Flores and J. Ortega, J. Chem. Phys., 2011, 134, 044701. 16 B. Pieczyrak, E. Abad, F. Flores and J. Ortega, J. Chem. Phys., 2011, 135, 084702. 17 F. Flores and J. Ortega, Basic theory of the Molecule-Metal interface, in The Molecule-Metal Interface, ed. N. Koch, N. Ueno and A. T. S. Wee, Wiley-VCH, 2013; F. Flores, E. Abad, J. I. Martı´nez, B. Pieczyrak and J. Ortega, J. Phys.: Condens. Matter, 2013, 25, 094007. 18 J. L. Bredas, J. E. Norton, J. Cornil and V. Coropceanu, Acc. Chem. Res., 2009, 42, 1691. 19 L. Muccioli, G. D’Avino and C. Zannoni, Adv. Mater., 2011, 23, 4532. ´das, Adv. Mater., 2013, 25, 878. 20 Y.-T. Fu, C. Risko and J.-L. Bre 21 J. P. Lewis, P. Jelı´nek, J. Ortega, A. A. Demkov, D. G. Trabada, B. Haycock, H. Wang, G. Adams, J. K. Tomfohr, E. Abad, H. Wang and D. A. Drabold, Phys. Status Solidi B, 2011, 248, 1989. 22 P. Jelı´nek, H. Wang, J. P. Lewis, O. F. Sankey and J. Ortega, Phys. Rev. B: Condens. Matter Mater. Phys., 2005, 71, 235101. 23 J. P. Lewis, K. R. Glaesemann, G. A. Voth, J. Fritsch, A. A. Demkov, J. Ortega and O. F. Sankey, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 64, 195103. 24 A. A. Demkov, J. Ortega, M. P. Grumbach and O. F. Sankey, Phys. Rev. B: Condens. Matter Mater. Phys., 1995, 52, 1618. 25 O. F. Sankey and D. Niklewski, Phys. Rev. B: Condens. Matter Mater. Phys., 1989, 40, 3979. 26 M. A. Basanta, Y. J. Dappe, P. Jelı´nek and J. Ortega, Comput. Mater. Sci., 2007, 39, 759. 27 M. Fuchs and M. Scheffler, Comput. Phys. Commun., 1999, 119, 67. 28 S. Grimme, J. Comput. Chem., 2006, 27, 1787. 29 Y. J. Dappe, J. Ortega and F. Flores, Phys. Rev. B: Condens. Matter Mater. Phys., 2009, 79, 165409; Y. J. Dappe, M. A. Basanta, F. Flores and J. Ortega, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 74, 205434. 30 C. C. Mattheus, Polymorphism and Electronic Properties of Pentacene, Rijksuniversiteit Groningen, 2002. 31 C. C. Mattheus, G. A. de Wijs, R. A. de Groot and T. T. M. Palstra, J. Am. Chem. Soc., 2003, 125, 6323.
Phys. Chem. Chem. Phys., 2014, 16, 4268--4274 | 4273
View Article Online
Published on 06 January 2014. Downloaded by UNIVERSITY OF ALABAMA AT BIRMINGHAM on 30/10/2014 03:57:20.
PCCP
32 P. A. Heiney, J. E. Fischer, A. R. McGhie, W. J. Romanow, A. M. Denenstein Jr., J. P. MacCauley, A. B. Smith and D. E. Cox, Phys. Rev. Lett., 1991, 66, 2911. 33 R. A. Cantrell, C. James and P. Clancy, Langmuir, 2011, 27, 9944–9954. 34 F. R. Manby and R. J. Knowles, J. Chem. Phys., 2000, 112, 7002. 35 F. Flores and J. Ortega, Basic theory of the Molecule-Metal interface, in The Molecule-Metal Interface, ed. N. Koch, N. Ueno and A. T. S. Wee, Wiley-VCH, 2013; F. Flores, E. Abad, J. I. Martı´nez, B. Pieczyrak and J. Ortega, J. Phys.: Condens. Matter, 2013, 25, 094007.
4274 | Phys. Chem. Chem. Phys., 2014, 16, 4268--4274
Paper
36 J. D. Sau, J. B. Neaton, H. J. Choi, S. G. Louie and M. L. Cohen, Phys. Rev. Lett., 2008, 101, 026804. 37 R. G. Endres, C. Y. Fong, L. H. Yang, G. Witte and Ch. Woll, Comput. Mater. Sci., 2004, 29, 362. 38 S. Kang, Y. Yi, C. Kim, S. Cho, M. Noh, K. Jeong and C. Whang, Synth. Met., 2006, 156, 32–37. 39 I. Salzmann, S. Duhm, R. Opitz, R. L. Johnson, J. P. Rabe and N. Koch, J. Appl. Phys., 2008, 104, 114518. 40 M. V. Madhava Rao, T.-S. Huang, Y.-K. Su and Y.-T. Huang, J. Electrochem. Soc., 2010, 157, H69. 41 H. Alves, A. S. Molinari, H. Xie and A. F. Morpurgo, Nat. Mater., 2008, 7, 574.
This journal is © the Owner Societies 2014
PCCP View Article Online
PAPER
Cite this: Phys. Chem. Chem. Phys., 2014, 16, 4268
View Journal | View Issue
The role of charge transfer in the energy level alignment at the pentacene/C60 interface J. Beltra´n,* F. Flores and J. Ortega Understanding the mechanism of energy level alignment at organic–organic interfaces is a crucial line of research to optimize applications in organic electronics. We address this problem for the C60–pentacene interface by performing local-orbital Density Functional Theory (DFT) calculations, including the effect of the charging energies on the energy gap of both organic materials. The results are analyzed within the
Received 27th November 2013, Accepted 25th December 2013
induced density of interface states (IDIS) model. We find that the induced interface potential is in the
DOI: 10.1039/c3cp55004d
mainly induced by the small, but non-negligible, charge transfer between the two compounds and the
range of 0.06–0.10 eV, in good agreement with the experimental evidence, and that such potential is multipolar contribution associated with pentacene. We also suggest that an appropriate external
www.rsc.org/pccp
intercompound potential could create an insulator–metal transition at the interface.
I. Introduction Organic electronics has emerged as a promising field of research devoted to improving the performance of devices such as light emitting diodes, photovoltaic cells or field effect transistors. In these multilayered organic-based devices a suitable tailoring of properties like high photocurrent quantum efficiencies can be achieved using different combinations of electron-donor and electron-acceptor layers.1–5 For such donor–acceptor heterojunctions some of the considered constituting materials are C60, whose mechanical, electrical and chemical properties are favorable for the fabrication of reliable devices,6 and pentacene (P5) which shows a high mobility of holes7 and distinct polymorph phases.8 It is also well-known that in organic electronics the role of the interface is crucial since key processes such as charge transfer or charge recombination occur in its vicinity. Consequently, a great effort has been made to understand the electronic properties of donor–acceptor interfaces and their related phenomena such as band bending and interface energy level alignment. The interface energy level alignment at organic/organic interfaces depends on the interface structure which is partially controlled by the growth process. Using appropriate methods like Molecular Beam Epitaxy, abrupt interfaces have been prepared and, in those cases, it has been found that the band alignment of the organic materials forming the junction is controlled by an interface induced potential,9,10 at variance with the Schottky–Mott limit in which the vacuum levels of both compounds are assumed to be aligned. Several models, like the Integer Charge Transfer11 and the Induced Density of
Interface States (IDIS),10 have been proposed to explain the formation of that potential. In particular, the IDIS-model relies on parameters like the charge neutrality level (CNL), which is an electronegativity marker for organic materials that predicts the amplitude and direction of charge transfer at the interface, and the interface screening parameter (S) which dictates how efficiently a shift in the CNL affects the induced potential.10 On the other hand, it has also been proposed recently by Linares et al.12 that, for a C60/P5 cluster, the level alignment shift for those materials is mainly provided by the multipole electrostatic potential created by the organic molecules. In this paper, we analyze the electronic properties and interface energy level alignment for the P5/C60 interface including in our DFT calculations the effects associated with the molecule charging energy that will allow us to correct appropriately the transport energy gap of both organic materials.13–17 In our calculations, we go beyond cluster models, as used in other studies,12,18 and consider periodic geometries that yield a better description of the long range potential created on the organic molecules by the charge rearrangement induced at the interface.13 This will allow us to make a detailed comparison between the effects on the interface energy level alignment due to the charge transfer or to the molecule multipole potentials. The P5/C60 interface geometry is analyzed in Section II. Then, in Section III we discuss our approach to analyze the energy level alignment at the interface between two organic materials using DFT calculations. Finally, in Section IV we present our results and conclusions.
II. Interface geometry ´rica de la Materia Condensada and Condensed Matter Physics Depto. de Fı´sica Teo Center (IFIMAC), Universidad Auto´noma de Madrid, 28049-Madrid, Spain. E-mail: [email protected]
4268 | Phys. Chem. Chem. Phys., 2014, 16, 4268--4274
Regarding the interface structure, predictions continue to be a challenging task12 with several studies19,20 suggesting a
This journal is © the Owner Societies 2014
View Article Online
Published on 06 January 2014. Downloaded by UNIVERSITY OF ALABAMA AT BIRMINGHAM on 30/10/2014 03:57:20.
Paper
disordered interface. In view of this lack of reliable information we have proceeded to calculate two different interface geometries, P5(001)/C60 and P5(011% )/C60, see Fig. 1 and 2. In the P5(001)/C60 interface, P5 is mainly oriented perpendicular to the surface, while in the P5(011% )/C60 interface P5 is almost parallel. These two interfaces represent two limiting cases for the P5 orientation, a fact that will allow us to suggest general conclusions even for disordered interfaces. Thus, we have calculated for each interface the most plausible geometry using DFT calculations that incorporate van der Waals interactions in a semi-empirical fashion. In these calculations we have used the local-orbital DFT code FIREBALL.21–24 FIREBALL uses real-space techniques and a basis set of numerical atomic-like orbitals (NAOs).25,27 This technique is based on a local-orbital formulation of DFT in which self-consistency is implemented on the orbital occupation numbers (see ref. 21 for details); in particular, in the present calculations the orbital occupation ¨wdin numbers have been obtained using the orthonormal Lo orbitals.24 We have used an optimized NAO basis set of s-(H) and sp3-(C) orbitals, with the following cut-off radii (in a.u.): s = 4.1 for H, and s = 4.5, p = 4.5 for C,26 and a k-space grid of 32 points homogeneously distributed in the 2-dimensional Brillouin zone. In these calculations we have used the Local Density Approximation (LDA)22 and the ion–electron interaction is modeled by means of norm-conserving pseudopotentials.27 In order to provide a more reliable geometry for the P5/C60 interface, a van der Waals interaction energy is included as an atom–atom attractive potential, fD(R)C6/R6, where R is the distance between each pair of atoms at the interface and fD(R) is the Grimme damping expression;14,28 C6 is taken as 14.5 eV A6 for the C–C interaction,29 which provides the relevant contribution to this energy.
Fig. 1 Adsorption geometry for the P5(011% )/C60 interface: top view and side view. The in-plane lattice vectors and the interface distance d are also shown. The two different C60-molecules are indicated (C60-1 and C60-2); and only the first P5 layer is plotted for simplicity.
This journal is © the Owner Societies 2014
PCCP
Fig. 2 Adsorption geometry for the P5(001)/C60 interface: top view and side view. The in-plane lattice vectors and the interface distance d are also shown.
It should be emphasized that, in order to avoid the double counting that might appear by combining simultaneously the LDA exchange–correlation energy and the van der Waals forces, we have introduced a kind of ‘‘corrected’’ LDA-approach13–15 whereby we eliminate the exchange–correlation contribution associated with the overlapping densities of both organic materials (see ref. 14 for more details). Then, the total interaction energy is calculated as the sum of the energies obtained using this ‘‘corrected’’ LDA-approach and the van der Waals forces as described above. P5 is known to have different polymorphs30 so its structure can vary depending on factors such as temperature or other growing conditions. We have considered the P5(001) and (011% ) surfaces at room temperature as given in references,30,31 and have combined each surface with a C60(001) structure with a distance between C60 neighboring centers approximately equal to 10 Å.32 Then, in order to obtain a plausible structure for the P5(011% )/C60 and the P5(001)/C60 interfaces, we have proceeded in the following way: (a) on the P5(011% ) surface structure, we have initially considered one C60 molecule in the unit cell (lattice vectors: a = 12.4 Å; b = 14.7 Å, see Fig. 1) letting the system relax until the atomic forces are smaller than 0.05 eV Å1. The obtained position, denoted C60-1 in Fig. 1, agrees well with one of the most stable configurations of a recent study.33 We then included a second C60 molecule in the unit cell (C60-2 in Fig. 1), relaxing all atomic positions again. We should stress that, with
Phys. Chem. Chem. Phys., 2014, 16, 4268--4274 | 4269
View Article Online
Published on 06 January 2014. Downloaded by UNIVERSITY OF ALABAMA AT BIRMINGHAM on 30/10/2014 03:57:20.
PCCP
Paper
the chosen P5 polymorph (see Fig. 1), the final C60–C60 distance at the interface is 9.6 Å, to be compared with 10 Å for the C60 bulk phase.32 Notice the different vertical distances of the two C60 positions on the top P5 surface: for the C60-1 molecule (blue circle in Fig. 1) the distance d from the layer of highest H atoms on P5 to the closest C atom in C60 is 1.7 Å, while for the C60-2 molecule that distance is 2.2 Å. (b) For the P5(001)–C60 interface, we only have one C60 molecule per unit cell; then, we again relax the C60 position until forces are smaller than 0.05 eV Å1, see Fig. 2. In this case the distance d is 2.0 Å, which is in-between the two values obtained for the P5(011% )/C60 interface. The in-plane lattice vectors in this interface form an angle of 102.31 and are 9.75 Å and 10.38 Å long while, as mentioned above, the C60–C60 distance in bulk C60 is 10.32
III. Calculation of energy levels at the organic/organic interface The theoretical analysis of the energy level alignment in organic/organic interfaces is a challenging task due the large size of these systems as well as the deficiencies of standard DFT techniques to properly describe the electronic structure of these organic interfaces.10 Due to the complexity of this analysis we use a practical and simplified approach13–15 based on local-orbital DFT (see Section II) in which appropriate corrections are introduced to take care of the main inaccuracies of the DFT calculation, see below. It is well known that in Kohn–Sham DFT calculations for organic semiconductors, local or semi-local exchange–correlation functionals give rise to small band gap values.10,17,34 In this respect, the critical role that different hybrid exchange–correlation functionals might play in the electronic properties of those materials at organic heterojunctions has been recently studied.35 In order to discuss how we have proceeded to tackle this problem at the P5/ C60 interface it is convenient to start presenting how accurate ionization and the affinity levels, eI and eA, for the isolated molecules can be determined from total energy DFT calculations for the neutral, anionic and cationic charged molecules respectively with N, N + 1 and N 1 electrons (D-SCF approach): eI = E [N] E [N 1] eA = E [N + 1] E [N],
(1)
where E [Ni] is the total energy for the ground state of the molecule with Ni electrons. The results of D-SCF calculations are usually in good agreement with the experimental evidence for the ionization and the affinity levels, eI and eA, for the isolated molecules. In contrast, the HOMO and LUMO Kohn– Sham levels obtained using local or semi-local exchange–correlation DFT functionals do not properly represent the transport gap for organic molecules. This is illustrated in Fig. 3. The experimental transport gap (ETg)0 = (eA eI), for C60 (P5) is 4.9 eV (5.2 eV) and the mid-gap is located 5.4 eV (4.0 eV) from the vacuum level. For comparison, the energy gaps obtained from converged36,37 DFT-LDA calculations, ELDA , for the neutral g molecules are: 1.6 eV (C60) and 1.2 eV (P5), very far from the
4270 | Phys. Chem. Chem. Phys., 2014, 16, 4268--4274
Fig. 3 (left) Energy level diagram showing the vacuum, highest occupied and lowest unoccupied levels as obtained from DFT Kohn–Sham LDA calculations for the isolated C60 and P5 molecules. (right) Experimental values of the affinity (eA) and ionization levels (eI) for the C60 and P5 molecules.
experimental values. This problem is related to the molecule self-interaction energy as described by the molecule charging energy U,10,17 i.e. (ETg)0 = ELDA + U. In Fig. 3 (left) we show the g LUMO and HOMO levels of the converged DFT-LDA calculations; notice also that the relative alignment of the DFT-LDA energy mid-gaps also differs from the ones obtained from experiment (or from a D-SCF calculation): while this difference is 0.8 eV in Fig. 3 (left), it amounts to 1.4 eV in Fig. 3 (right). The energy levels of the isolated molecules are shifted at the interface due to the interaction with the other molecules. The effect of this environment is threefold:17 (A) broadening of the energy levels into bands; (B) relative shift of the C60 and P5 molecular levels due to electrostatic and Pauli exclusion effects; (C) ‘‘dynamical’’ polarization or screening effects, shifting in different directions of occupied and empty states. Points (A) and (B) are obtained from our DFT calculations; in particular point (B) gives rise to an interface potential between both crystals. The effect of point (C) is a reduction of the transport gaps at the interface (or in the crystals) as compared with the values for isolated C60 or P5 molecules: ETg = (ETg)0 dU. Moreover, this gap reduction is also reflected in a shift of the ionization and affinity levels given by dU/2 and dU/2, respectively.10,17 The effects of the self-interaction energy and dynamical polarization response on the interface electronic structure are included in our calculations in a practical and simplified way introducing for each molecule a scissor operator:10 Ua X Oscissor ¼ (2) fjmi ihmi j jn i ihn i jg; a 2 ðmnÞ |mii and |nii being the empty (occupied) orbitals of the isolated molecule; Ua is the charging energy of the a-molecule. Notice that in eqn (2): U = (ETg)0 ELDA dU for each molecule a. We g also introduce for each molecule a rigid shift of the molecular levels by means of a shift operator: X Oshift ¼ ea (3) jbihbj; a ðbÞ
This journal is © the Owner Societies 2014
View Article Online
Published on 06 January 2014. Downloaded by UNIVERSITY OF ALABAMA AT BIRMINGHAM on 30/10/2014 03:57:20.
Paper
|bi being the orbital for each isolated molecule. Using these operators, we can fix in our local-orbital DFT calculations the initial value of the HOMO–LUMO gap and relative mid-gap position for the P5 and C60 molecules. Still, we have to determine the appropriate values of Ua and ea for each molecule at the P5/C60 interface. For this, instead of trying an ab initio approach we have resorted to using the experimental evidence.38–40 Several experimental studies38–40 have reported energy gaps for C60 and P5, respectively of 2.6 and 2.2 eV, at the P5/C60 interface. It is worth mentioning that theoretical calculations for the C60/Au(111) and P5/Au(111) interfaces14,16 have yielded energy gaps of 3.1 eV (C60) and 2.7 eV (P5). Notice that in both cases the energy gap for C60 is 0.4 eV larger than that for P5, the difference being that the experimental gaps for the P5/C60 interface are 0.5 eV smaller than the theoretical ones for the Au(111)/organic contact. Thus, the values ETg(C60) = 2.6 eV and ETg(P5) = 2.2 eV seem very reasonable. This information is completed with the values of the mid-gap positions for the P5 and C60 molecules, which are obtained from Fig. 3 (right). Thus, we use the scissor and shift operators, eqn (2) and (3), to adjust the initial values of the energy gaps and mid-gap positions for each molecule; then, our DFT calculation is performed after the operators, Oshift and a Oscissor , are introduced as a kind of pseudopotential. a
IV. Results and discussion Fig. 4 shows the energy level schemes for both the P5(011% )/C60 and P5(001)/C60 interfaces after electronic self-consistency. As shown in this figure, the interface induced potential at the P5(011% )/C60 interface is 0.14 eV, while for the P5(001)/C60 interface is 0.04 eV. The average of these two values, 0.09 eV, can be taken as a good representation of the interface potential for a disordered interface and agrees well with the experimental interface potential ranging between 0.07 eV and 0.15 eV.11,38,39
Fig. 4 Energy level diagrams of the vacuum, highest occupied and lowest unoccupied levels for the P5(011% )/C60 (left) and P5(001)/C60 (right) interfaces after electronic selfconsistency. These diagrams correspond to the D = 0 case (see text).
This journal is © the Owner Societies 2014
PCCP
For the (LUMOC60 – HOMOP5) difference we find 1.1 eV at the P5(011% )/C60 interface and 0.9 eV at the P5(001)/C60 interface, which should be compared with reported experimental values between 1.6 and 1.1 eV.11,38,39 A good understanding of the mechanism behind the interface induced potential can be reached by analyzing how the interface properties depend on the initial relative alignment between the mid-gaps of both materials; this would correspond to applying an external potential (D), or a rigid shift to, say, the P5 energy levels with respect to the C60 levels. This shift is introduced in our calculations using the Oshift operator mentioned a above, see eqn (2). In this way we can analyze the induced potential, the charge transfer and the (LUMOC60 – HOMOP5) evolution as a function of D. Notice that the cases shown in Fig. 4 correspond to D = 0 eV. Fig. 5 shows, for the P5(011% )/C60 and P5(001)/C60 interfaces, the (LUMOC60 – HOMOP5) and the charge transfer per molecule as a function of D (positive values of D means that the P5 levels are displaced upwards). For negative values of D, (LUMOC60 – HOMOP5) increases and the charge transfer is small. However, for positive values of D the difference (LUMOC60 – HOMOP5) decreases and the charge transfer from P5 to C60 increases, indicating that this charge transfer is associated with the interaction between the LUMOC60 and HOMOP5 levels. Interestingly, for large values of D (D > 0) the difference between the LUMOC60 and the HOMOP5 level for the P5(011% )/C60 and P5(001)/C60 interfaces converges to 0.50 and 0.25 eV respectively. This is related to the onset of a strong overlap between the LUMOC60 and HOMOP5 states, which is also reflected in the
Fig. 5 (LUMOC60 – HOMOP5) difference and charge transfer per molecule at the P5(011% )/C60 and P5(001)/C60 interfaces as a function of D (see text). The estimated charge error is around 0.005e.
Phys. Chem. Chem. Phys., 2014, 16, 4268--4274 | 4271
View Article Online
Published on 06 January 2014. Downloaded by UNIVERSITY OF ALABAMA AT BIRMINGHAM on 30/10/2014 03:57:20.
PCCP
Paper
Fig. 6 Density of states projected on the P5 or C60 organic materials for the P5(011% )/C60 interface. We have used a Lorentzian broadening of 0.05 eV for the sharp states. Left and right plots correspond to D = 0 and D = 1.5 eV respectively; notice the overlap between the HOMOP5 and the LUMOC60 levels in the latter.
significant increase of charge transfer as D increases further. This strong overlap gives rise to a metallic interface. Fig. 6 (left) shows the density of states for the P5(011% )/C60 interface projected onto either the P5 or C60 molecules, for D = 0. In this case the HOMOP5 level is around 1 eV below the LUMOC60 level and the charge transfer is small (see Fig. 5). Hybridization features, such as interfacial states corresponding to the interaction between P5 and C60, can be seen at the middle of the P5 band gap. For comparison, Fig. 6 (right) shows the density of states for D = +1.5 eV, a case presenting a metallic interface due to the strong overlap between the HOMOP5 and the LUMOC60 levels. These results show that the charge transfer plays an important role in the level alignment between the C60 and P5 organic layers. In particular, the potential induced between the organic compounds, Vinduced, follows the trend of the charge transfer. Fig. 7 shows this induced interface potential, Vinduced, as a function of D; notice how Vinduced increases for positive values of D as corresponds to a larger charge transfer between both materials, showing also a change of trend to a metallic interface for values of D around 0.5 eV. It is interesting to point out that for D > 1 eV the initial position of the P5 HOMO level is above the C60 LUMO level, suggesting a transfer of charge of 1 electron from P5 to C60. The transfer of charge between the P5 and C60 layers creates, however, a large long-range induced dipole potential between both layers (Fig. 7), shifting downwards the P5 levels w.r.t. the C60 levels. The final values for the P5 and C60 levels and charge transfer are obtained from the selfconsistent solution of this problem (Fig. 5). For example, for D = 2 eV this yields a charge transfer of 0.15–0.35 electrons per molecule and a final position for the P5 HOMO level 0.25–0.5 eV below the C60 LUMO level (Fig. 5). On the other hand, for negative values of D, Vinduced decreases quite slowly, as corresponds to an insulating interface. This situation can be described by means of the screening parameter S35 that can dVinduced (for a full insulating dD interface S = 1, i.e. no screening of the external potential D, while S = 0 for a completely metallic case). For the insulating be obtained from Fig. 7 as S ¼ 1
4272 | Phys. Chem. Chem. Phys., 2014, 16, 4268--4274
Fig. 7 Induced interface potential, Vinduced, as a function of D for the P5(011% )/C60 and P5(001)/C60 interfaces.
regime we find S = 0.96 0.02 while for the metallic one S = 0.03 0.01, this last small value reflecting the high screening properties associated with the metallic interface. We also find that the charge transfer vanishes (see Fig. 5) for D C 1.6 or +0.4 eV for the P5(011% )/C60 and P5(001)/C60 interfaces, respectively. For these values of D we find that Vinduced = 0.025 or 0.065 eV for the P5(011% )/ C60 and P5(001)/C60 interfaces, respectively, but not zero, indicating that another mechanism different from the charge transfer is operating here: as shown by Linares et al.12 this is associated with the potential created by the multipoles of the neutral P5 molecules, a mechanism responsible for the interface potential appearing at zero charge transfer. Therefore, in our calculations we find that for the P5(011% )/C60 interface, in the case D = 0, Vinduced = 0.14 eV while the multipolar contribution is 0.025 eV. However, for the P5(001)/ C60 interface (D = 0), Vinduced = 0.04 eV, the multipolar contribution being as large as 0.065 eV. Taking the average of these two interfaces as a reasonable representation of a disordered interface, we find that Vinduced C 0.09 eV, while the multipolar potential is B0.045 eV. We conclude that, in addition to the charge transfer mechanism, the multipolar term yields an important contribution to the total induced interface potential for D = 0, in agreement with ref. 12. Finally, we should stress that, as commented above, for large positive values of D we find a strong overlap between the
This journal is © the Owner Societies 2014
View Article Online
Published on 06 January 2014. Downloaded by UNIVERSITY OF ALABAMA AT BIRMINGHAM on 30/10/2014 03:57:20.
Paper
acceptor’s LUMOC60 and donor’s HOMOP5 levels, making the interface to have a metallic behavior and to show a pinning of the (LUMOC60 – HOMOP5) value. This indicates that a metal–insulator transition appears at this interface for D B 0.5 eV, associated with the overlapping of the LUMOC60 and the HOMOP5 levels (see Fig. 6 (right)). We also mention that a similar transition has been recently reported for the TCNQ/TTF interface;14,41 in that case, however, the TCNQ/TTF interface is already metallic for D = 0. For the P5–C60 case, we suggest that an appropriate external potential between the two organic materials could create that kind of metal–insulator phase transition. In conclusion, we have analyzed the energy level alignment at the interface between the C60 and pentacene organic semiconductors using a DFT approach with corrections to properly describe the energy gaps of the organic materials. We have studied two different interface geometries, corresponding to the (001) and (011% ) surfaces of pentacene. We find that the interface potential induced at the P5/C60 interface is mainly due to two effects: the charge transfer between the two materials and the multipolar contribution associated with pentacene. For a disordered P5/C60 interface we find, after averaging over the values calculated for the P5(011% )/C60 and P5(001)/C60 interfaces, that the total induced interface potential is B0.09 eV, while the multipolar contribution to that induced interface potential is found to be B0.045 eV. We have also shown that, depending on an effective external potential acting between the C60 and P5 organic materials, the interface can show either an insulating or a metallic character.
Acknowledgements This work is supported by the Spanish MICIIN under contract FIS2010-16046, the CAM under contract S2009/MAT-1467, the European Project MINOTOR (Grant FP7-NMP-228424) and COST-CMTS Action CM1002 (CODECS).
References 1 R. Cantrell and P. Clancy, Surf. Sci., 2008, 602, 3499–3505. 2 S. Duhm, G. Heimel, I. Salzmann, H. Glowatzki, R. L. Johnson, A. Vollmer, J. P. Rabe and N. Koch, Nat. Mater., 2008, 7, 326. 3 T. Hosokai, N. Mitsuo, S. Noro, T. Nakamura, S. Kera, K. Sakamoto and N. Ueno, Chem. Phys. Lett., 2010, 487, 67. 4 K. Akaike, K. Kanai, Y. Ouchi and K. Seki, Adv. Funct. Mater., 2010, 20, 715. 5 N. Hiroshiba, R. Hayakawa, T. Chikyow, Y. Yamashita, H. Yoshikawa, K. Kobayashi, K. Morimoto, K. Matsuishi and Y. Wakayama, Phys. Chem. Chem. Phys., 2011, 13, 6280. 6 D. Dougherty, W. Jin, W. Cullen, G. Dutton, J. Reutt-Robey and S. Robey, Phys. Rev. B: Condens. Matter Mater. Phys., 2008, 77, 073414. 7 K. Kim, Y. K. Yoon, M. Mun, S. P. Park, S. S. Kim, S. Im and J. H. Kim, J. Supercond., 2002, 15, 595–598. 8 C. C. Mattheus, G. a. de Wijs, R. a. de Groot and T. T. M. Palstra, J. Am. Chem. Soc., 2003, 125, 6323–6330.
This journal is © the Owner Societies 2014
PCCP
9 H. Ishii, K. Sugiyama, E. Ito and K. Seki, Adv. Mater., 1999, 11, 605; I. G. Hill, et al., Appl. Surf. Sci., 2000, 166, 354. ´zquez, Phys. Chem. Chem. Phys., 10 F. Flores, J. Ortega and H. Va 2009, 11, 8658–8675. 11 X. Liu, Y. Zhan, S. Braun, F. Li and M. Fahlman, Phys. Rev. B: Condens. Matter Mater. Phys., 2009, 80, 115401; S. Braun, W. R. Salaneck and M. Fahlman, Adv. Mater., 2009, 21, 1450. 12 M. Linares, D. Beljonne, J. Cornil, K. Lancaster, J. L. Bredas, S. Verlaak, A. Mityashin, P. Heremans, A. Fuchs, C. Lennartz, P. Aurel and L. Ducasse, J. Phys. Chem. C, 2010, 114, 3215–3224. ´n, F. Flores, J. I. Martı´nez and J. Ortega, J. Phys. 13 J. I. Beltra Chem. C, 2013, 117(8), 3888. 14 E. Abad, J. Ortega and F. Flores, Phys. Status Solidi A, 2012, 209, 636; E. Abad, J. Ortega, Y. J. Dappe and F. Flores, Appl. Phys. A: Mater. Sci. Process., 2009, 95, 119–124. 15 E. Abad, Y. J. Dappe, J. I. Martı´nez, F. Flores and J. Ortega, J. Chem. Phys., 2011, 134, 044701. 16 B. Pieczyrak, E. Abad, F. Flores and J. Ortega, J. Chem. Phys., 2011, 135, 084702. 17 F. Flores and J. Ortega, Basic theory of the Molecule-Metal interface, in The Molecule-Metal Interface, ed. N. Koch, N. Ueno and A. T. S. Wee, Wiley-VCH, 2013; F. Flores, E. Abad, J. I. Martı´nez, B. Pieczyrak and J. Ortega, J. Phys.: Condens. Matter, 2013, 25, 094007. 18 J. L. Bredas, J. E. Norton, J. Cornil and V. Coropceanu, Acc. Chem. Res., 2009, 42, 1691. 19 L. Muccioli, G. D’Avino and C. Zannoni, Adv. Mater., 2011, 23, 4532. ´das, Adv. Mater., 2013, 25, 878. 20 Y.-T. Fu, C. Risko and J.-L. Bre 21 J. P. Lewis, P. Jelı´nek, J. Ortega, A. A. Demkov, D. G. Trabada, B. Haycock, H. Wang, G. Adams, J. K. Tomfohr, E. Abad, H. Wang and D. A. Drabold, Phys. Status Solidi B, 2011, 248, 1989. 22 P. Jelı´nek, H. Wang, J. P. Lewis, O. F. Sankey and J. Ortega, Phys. Rev. B: Condens. Matter Mater. Phys., 2005, 71, 235101. 23 J. P. Lewis, K. R. Glaesemann, G. A. Voth, J. Fritsch, A. A. Demkov, J. Ortega and O. F. Sankey, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 64, 195103. 24 A. A. Demkov, J. Ortega, M. P. Grumbach and O. F. Sankey, Phys. Rev. B: Condens. Matter Mater. Phys., 1995, 52, 1618. 25 O. F. Sankey and D. Niklewski, Phys. Rev. B: Condens. Matter Mater. Phys., 1989, 40, 3979. 26 M. A. Basanta, Y. J. Dappe, P. Jelı´nek and J. Ortega, Comput. Mater. Sci., 2007, 39, 759. 27 M. Fuchs and M. Scheffler, Comput. Phys. Commun., 1999, 119, 67. 28 S. Grimme, J. Comput. Chem., 2006, 27, 1787. 29 Y. J. Dappe, J. Ortega and F. Flores, Phys. Rev. B: Condens. Matter Mater. Phys., 2009, 79, 165409; Y. J. Dappe, M. A. Basanta, F. Flores and J. Ortega, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 74, 205434. 30 C. C. Mattheus, Polymorphism and Electronic Properties of Pentacene, Rijksuniversiteit Groningen, 2002. 31 C. C. Mattheus, G. A. de Wijs, R. A. de Groot and T. T. M. Palstra, J. Am. Chem. Soc., 2003, 125, 6323.
Phys. Chem. Chem. Phys., 2014, 16, 4268--4274 | 4273
View Article Online
Published on 06 January 2014. Downloaded by UNIVERSITY OF ALABAMA AT BIRMINGHAM on 30/10/2014 03:57:20.
PCCP
32 P. A. Heiney, J. E. Fischer, A. R. McGhie, W. J. Romanow, A. M. Denenstein Jr., J. P. MacCauley, A. B. Smith and D. E. Cox, Phys. Rev. Lett., 1991, 66, 2911. 33 R. A. Cantrell, C. James and P. Clancy, Langmuir, 2011, 27, 9944–9954. 34 F. R. Manby and R. J. Knowles, J. Chem. Phys., 2000, 112, 7002. 35 F. Flores and J. Ortega, Basic theory of the Molecule-Metal interface, in The Molecule-Metal Interface, ed. N. Koch, N. Ueno and A. T. S. Wee, Wiley-VCH, 2013; F. Flores, E. Abad, J. I. Martı´nez, B. Pieczyrak and J. Ortega, J. Phys.: Condens. Matter, 2013, 25, 094007.
4274 | Phys. Chem. Chem. Phys., 2014, 16, 4268--4274
Paper
36 J. D. Sau, J. B. Neaton, H. J. Choi, S. G. Louie and M. L. Cohen, Phys. Rev. Lett., 2008, 101, 026804. 37 R. G. Endres, C. Y. Fong, L. H. Yang, G. Witte and Ch. Woll, Comput. Mater. Sci., 2004, 29, 362. 38 S. Kang, Y. Yi, C. Kim, S. Cho, M. Noh, K. Jeong and C. Whang, Synth. Met., 2006, 156, 32–37. 39 I. Salzmann, S. Duhm, R. Opitz, R. L. Johnson, J. P. Rabe and N. Koch, J. Appl. Phys., 2008, 104, 114518. 40 M. V. Madhava Rao, T.-S. Huang, Y.-K. Su and Y.-T. Huang, J. Electrochem. Soc., 2010, 157, H69. 41 H. Alves, A. S. Molinari, H. Xie and A. F. Morpurgo, Nat. Mater., 2008, 7, 574.
This journal is © the Owner Societies 2014
